A novel fusion method of infrared and visual images based on Undecimated Discrete Wavelet Transform and saliency map

This paper proposes a new fusion approach of infrared (IR) and visual (VIS) images based on Undecimated Discrete Wavelet Transform (UDWT) and saliency map analysis. The proposed fusion method is composed of three steps. Firstly, IR and VIS images are decomposed into low-pass sub-band coefficients and high-pass sub-band coefficients by UDWT. Meanwhile, sub-band directional coefficients are extracted by Non-Subsampled Directional Filter Banks (NSDFB) from high-pass sub-band coefficients. Secondly, the coefficient combination rules are set based on saliency maps of original IR and VIS images, which are obtained by employing the Itti saliency computing model. Thirdly, fused images are re-built by the inverse UDWT. The results of the experiments indicate that the presented approach achieves better subjective fusion effects than some current popular fusion methods. Furthermore, the proposed method gets better objective assessments.